Fix four things a 0.116 readiness review reproduced - #157
Merged
guanzhousongmicrosoft merged 1 commit intoAug 25, 2026
Merged
Conversation
Each of these was found by running the site's own instructions, not by reading them, so each fix is stated in the terms the failure appeared in. Pinning a version fails as documented. The Download page derives one version string from the extension packages and prints it everywhere, but the two package families are versioned differently: `postgresql-18-documentdb` is `0.116-0` while the full-stack `documentdb-18` is `0.116.0`. Following the page gives `E: Version '0.116-0' for 'documentdb-18' was not found`. Derive both, name both, and say which belongs to which -- and stop hardcoding four package names in the pin commands so they track the actual selection. The Docker quick start produced an empty database. `--init-data` defaults to false, and `--skip-init-data` is a legacy alias for `--init-data false`, so the documented command yields `postgres, template0, template1` and a sample collection count of 0 while the page claimed sample data loads by default. Pass `--init-data true` and correct the table. Also add the image tag list: `latest` is `pg17`, and the tags carry the PostgreSQL major, which is the only way to evaluate on the same major you intend to deploy on. Two doc URLs were dead and nobody could see it. `redirect()` cannot work under `output: "export"` -- there is no server to send a 3xx -- so Next rendered both as `<html id="__next_error__">` bodies served with HTTP 200. They were in sitemap.xml and linked from two live pages, and every status-code link checker called them healthy. Render a real page that links to the replacement and forwards on hydration, mark it noindex so the sitemap step drops it, and fail the build if any route ever ships as an error document again. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 8df9084a-ccaf-432c-b015-2ccd8893a9d8 Signed-off-by: Guanzhou Song <guanzhousong@users.noreply.github.com>
guanzhousongmicrosoft
approved these changes
Aug 25, 2026
guanzhousongmicrosoft
left a comment
guanzhousongmicrosoft
left a comment
Contributor
There was a problem hiding this comment.
Verified each failure and each fix locally: the pin command's E: Version not found, the empty database from the documented Docker run, and the two 47/61-byte HTTP 200 bodies. Build guard, sitemap output and rendered HTML all check out.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A readiness review ran the site's own instructions against real packages and
real containers instead of reading them. Four things failed. Each fix below is
described in the terms the failure actually appeared in.
Pinning a version fails exactly as documented
The Download page derives one version string from the extension packages and
prints it in every command. But the two package families are versioned
differently:
So the documented pin command fails:
releaseInfonow derives both families, and the page names both and says whichbelongs to which. The pin commands also stopped hardcoding four package names,
so they follow whatever the reader actually selected.
The Docker quick start produced an empty database
Running the documented command verbatim leaves you with
postgres,template0and
template1, and a sample collection count of0— while the page saidsample data loads by default.
--init-datadefaults to false, and--skip-init-datais a legacy alias for--init-data false.The command now passes
--init-data true, the claim is gone, and the optiontable matches the binary.
Related, on the same page: the image tags are now listed.
latestispg17(identical manifest hash), and the tags carry the PostgreSQL major. Without
that, someone evaluates on 17 and then deploys the
documentdbmeta package,which pulls PG 18 — a silent major version change between the two steps.
Two documentation URLs were dead, and nothing could see it
/docs/getting-started/prebuilt-packages/and/docs/getting-started/vscode-extension-guide/served 47 and 61 byte bodies.The cause is that
redirect()cannot work underoutput: "export"— there isno server to send a 3xx — so Next renders the thrown redirect as an error
document. Both were listed in
sitemap.xmland linked from two live pages, andbecause GitHub Pages returns HTTP 200 for them, every status-code link
checker reported them healthy.
They now render a real page linking to the replacement, forward on hydration,
and are marked
noindexso the sitemap step drops them.And a guard so this particular failure can't come back silently
generate-sitemap.mjsnow detects<html id="__next_error__">while it walksout/and fails the build with the offending URLs. This is the check thatwould have caught the above; a status-code check never could.
Verification
tsc --noEmit,eslint,vitest(130 tests) — all cleannext build+ sitemap: 274 URLs, 4 noindex skipped, guard exit 0, and bothmoved routes are gone from
sitemap.xml/packages, the pinwarning,
--init-data trueon the Docker page, the sample-data claim gone,and
pg17-0.116.0/version.txt/--skip-init-dataon documentdb-localDeliberately not changed
The review also reported that
Admin100no longer exists and the containerrefuses to start without
--password. Running it says otherwise:It starts, and only warns that a future release will refuse.
--helpsaysREQUIRED.while the runtime defaults — that's a product inconsistency to fixin the container, not on the website. The table here is currently accurate.